from paris with love

Mimoune,

I’m not sure if I understand the paper correctly, I’m looking for any mention of the settings used for the reference data (how the reference durations were derived). Is there any?

It was mentioned that:

"For none of the 1560 instances,
any PMIS computed a schedule with a shorter project duration than the reference value."

and

"In this paper, we have experimentally analyzed the resource-allocation capabilities of
seven commercial PMIS. For the resource- and precedence-constrained project
scheduling problem RCPSP, it has turned out that when using any of these packages for
resource allocation, the project duration obtained is noticeably longer than necessary."

How did they arrived at the "necessary duration(s)" (using the reference data)?

Are the settings used for comparison between the reference data and the subject softwares were the same?

They only mentioned that they used "default settings" and "best options" for the softwares to calculate durations.

Can anybody clarify?

P6 default algorithms are awful.
It is not easy to check their conclusions because the set of sample projects was not defined.
It would be interesting to try Spider with this set.
Thank you for the reference.
Vladimir

Vladimir and anoon

the data set, the optimum solution (using state of the art algorithm) are downloaded from
http://129.187.106.231/psplib/


and Vladimir yes you can test spider with this, who knows perhaps spider will give a shorter duration then this academic algorithm

best regards

Hi Mimoune,
I tried but could not understand the format.
Unfortunately academic algorithms usually do not work with large dimensions.

Best Regards,
Vladimir

Vladimir

i am sure you have already read this paper,
http://129.187.106.231/psplib/files/KH-18-2-05.pdf
it is very interesting to note the recent academic research in this field, i bet it will not take a lot of time before scheduling software will use those new approach.

Vladimir,(if u don’t mind) the algorithm used by spider can be classified in which category ;)

best regards

The closest is FBI though our iterative algorithms are more complicated and are used since first launch of Spider in 1993. So for us "new approach" is not the right word.
Best Regards,
Vladimir